Microelectronics Reliability 39 (1999) 14611472

www.elsevier.com/locate/microrel

Impact of power electronics packaging on the reliability of grid connected photovoltaic converters for outdoor applications
Mathu M. Meinhardt, V. Leonavicius*, J. Flannery, S.C. O na
PEI Technologies, National Microelectronics Research Centre, University College, Lee Maltings, Prospect Row, Cork, Ireland Received 30 April 1999; received in revised form 25 June 1999

Abstract This paper deals with thermal and reliability aspects of converters for grid connected photovoltaic applications (rate power approximately 100 W) which can be integrated into solar modules. The use of these Module Integrated Converters (MIC) promises a reduction of costs due to mass production. It improves behaviour of the whole photovoltaic system, as there is no voltage mismatch caused by shading eects. The electronic components of the MIC are exposed to extreme environmental conditions. Lifetime limiting factors are described. Methods to increase the lifetime of the most crucial components in the MIC, electrolytic capacitors, are described. The presented 1st generation Low Prole MIC with ``optimised design'' is capable of running in very high ambient temperatures. This design maximises availability of the photovoltaic system, which consequently leads to a cost reduction of the electrical energy delivered to the grid. Dierent realisations of how to integrate the designed Low Prole MIC into the solar module are discussed. A detailed thermal simulation is used to optimise the MIC design according to temperature and reliability issues. The models used for thermal simulation are described. # 1999 Elsevier Science Ltd. All rights reserved.
Keywords: Power electronics packaging; Converter; Thermal simulations; Integration; Miniaturisation; Failure rate; Market analysis; Environmental tests; Switched mode power supply

1. Introduction The increased use of renewable energy and rational use of conventional energy sources are fundamental pillars of a responsible energy policy for the future. Because of their sustainable character, renewable energy sources almost exclusively based on solar

* Corresponding author. Tel.: +353-21-904241; fax: +35321-270271. E-mail address: vleon@nmrc.ucc.ie (V. Leonavicius)

energy are capable of preserving resources and providing energy services virtually without any environmental impact. Grid connected Photovoltaic (PV) systems distinguish themselves in the method of energy transfer: there is no need for energy storage devices such as batteries. Recent trends in grid connected photovoltaic systems show coexistence of dierent kinds of converter concepts. The combination of Module Integrated Converter (MIC) and solar module often called ``AC Solar Module'' allows a direct connection of the solar modules to the grid by converting the module

0026-2714/99/$ - see front matter # 1999 Elsevier Science Ltd. All rights reserved. PII: S 0 0 2 6 - 2 7 1 4 ( 9 9 ) 0 0 0 8 0 - 3

1462

M. Meinhardt et al. / Microelectronics Reliability 39 (1999) 14611472

Fig. 1. Cross-section of a MIC mounted to the frame of a solar module.

The market study shows that worldwide sales of solar cells have been increasing steadily over the last few years [1]. Fig. 2 shows an optimistic and a pessimistic forecast for the development of the PV market. According to the most pessimistic forecast, the PV market in 2005 will be about 400 MWp worldwide. The estimation about the market shares of dierent applications of solar cells is presented in Fig. 3. Assuming that 17% of the worlds solar cell production will be used in AC modules (small or medium scale grid connected systems), MICs with cumulative power of 68 MWp could be sold (again, according to the pessimistic forecast). Considering that the rated power of one MIC is in a range of 100300 Wp, this market volume corresponds to 230,000680,000 pieces per year in 2005.

DC voltage to a 230 V AC voltage (Fig. 1). MICs have some important advantages over string orientated or central inverter concepts such as: . increased energy yield in case of systems suering from shading eects; . reduced danger of arcs due to replacement of the DC installation by an AC installation; . simplied conguration of PV systems due to standardised interface (230 V AC). Usually the rated power of the MIC is slightly lower than the rated power of the solar module. In the future a large number of solar modules will have an average power of 100150 W which corresponds to an area of solar cells of 11.5 m2. Depending on the PV system's site and properties of installation, string oriented systems are likely to be used in the medium power range. Nevertheless, the central inverter concept will remain popular, especially in very large-scale PV systems.

2. Status of reliability of the MIC Recent technology and design is already capable of guaranteeing the lifetime of solar modules for 1520 years of continuous operation. Being a part of the whole PV system, the MIC should have similar lifetime for obvious reasons. In order to maximise the energy yield of the whole PV system all parts should be equally reliable. Economical and environmental aspects play a very important role in it, for the production company as well as for the user. Manufacturers must guarantee a minimum lifetime of 15 years for the MIC to win market share and reputation. Unfortunately, they state just expected lifetime values so far. Calculations based on statistical models (i.e. handbook [2]) represent just a very rough theoretical value. Some calculations based on accelerated lifetime tests cover particular aspects like high temperature and humidity. The electronic components of the MIC are exposed to extreme environmental conditions. High thermo-mechanical stress has a very large impact on the lifetime of the MIC. But in practice, there are many more factors

Fig. 2. Worldwide PV market increase and estimated market for MIC [1].

Fig. 3. Estimated market potential for MIC.

M. Meinhardt et al. / Microelectronics Reliability 39 (1999) 14611472

1463

Fig. 4. MIC lifetime limiting factors.

(Fig. 4) that have a complex impact. This shows that the reliability issues should be treated seriously. The demonstrated lifetime (i.e. the real value) as opposed to the calculated lifetime is the most critical factor that should be addressed. However, it takes a long time to obtain this value (i.e. hours of operation) and thus not practical. Failures can be categorised into design related failures and those which are triggered by operational conditions (environmental impact). In most cases failures are caused by a combination of both factors. Design related factors are: selection of components and materials, number, rating and physical placement of components. Failures related to operational conditions of the MIC are either of deterministic or unpredictable nature. The maximum absolute temperature, the number and amplitude of temperature cycles as well as humidity are very much predictable. Whereas the elec-

trical disturbances coming from the mains or in the case of lightning strikes on the solar module are less predictable since they are caused by the weather or other stochastic events in the electric grid (e.g. failures in other appliances connected to the same grid). Failure rate related with design and predictable environmental factors can be reduced by using advanced packaging techniques. The crucial components in terms of the highest failure rate are the aluminium electrolytic capacitors of the DC-lter and the Mosfets [3]. DC electrolytic capacitors cause more than 50% of all MIC failures. Increased lifetime of the DC capacitors would signicantly improve the reliability of the whole MIC. The lifetime of aluminium electrolytic capacitors is inuenced mainly by these three parameters: . internal temperature (hot spot temperature) of the capacitor; . voltage stress (ratio of applied voltage to rated voltage); . current stress (ratio of applied ripple current to rated ripple current). The impact of temperature and voltage stress on the lifetime of aluminium electrolytic capacitors is illustrated in Fig. 5a and b. The internal temperature of aluminium electrolytic capacitors is the main factor which has the biggest inuence on their lifetime. Therefore the lifetime can be increased by reducing the operating temperature (case temperature) or using capacitors with higher rated temperature. As shown in Fig. 5a, the lifetime of the DC-lter operating at 958C can be increased by a factor of three if a 1258C capaci-

Fig. 5. Inuence of temperature and voltage stress on the lifetime of aluminium electrolytic capacitors.

1464

M. Meinhardt et al. / Microelectronics Reliability 39 (1999) 14611472

Fig. 6. Power density and height of Low Prole MIC, commercially available MICs and power supplies.

tor is used instead of a 1058C capacitor. But the size of the DC-lter would then be doubled (increased cost). The voltage stress (Fig. 5b) of the capacitors also has a big inuence on their lifetime. According to the publications [2,4,5], lifetime can be increased by a factor of four to 30 by limiting the capacitor's voltage stress to 0.5. Reduction of the current stress increases lifetime up to 25% [4] because of lower internal losses and lower hot spot temperature of the capacitors. The disadvantage of all three possibilities is that they lead to higher costs and increased size of the DClter and the whole MIC. The best way to extend the lifetime is achieved by decreasing the internal temperature using a proper thermal design of the entire MIC.

3. Power electronics packaging of the MIC The results of a market survey of a MIC and conventional power supplies (AC/DC and DC/DC) illustrated in Fig. 6 show that nearly all commercial converters are greater in height than the frame of most solar modules. The power density of power supplies is commonly used as a general measure of technical advancement. The power density of the most MICs is also small compared to AC/DC or DC/DC power supplies. Therefore miniaturisation and cost reduction are required. However, decreased size and higher power density require better thermal management. These conicting aspects show the challenge for the power electronics packaging to nd a compromise between cost, miniaturisation and reliability. Dierent design possibilities were evaluated using thermal simulation and failure analysis. Following this procedure, an optimum design with lower failure rate was found. It is proposed to target a Low Prole MIC design. In this case, thermal resistance of any point in

the converter is lower. Simulations of the simplied thermal model [3] shows that a decreased height of a MIC leads to reduced component temperatures and consequently to a higher reliability of the MIC. The main drawback of this design, however, is the requirement of the low prole electronic components. That means restricted selection of manufacturers and therefore potentially higher costs. But on the other hand, it is possible to reduce the costs if the whole size of the MIC is reduced (increased density). And this can be achieved by means of advanced design and power packaging techniques such as application of planar devices, integration of the electronic components, encapsulation of the converter. As a result, a smaller amount of materials is required for the production of the MIC, for example potting material for encapsulation. Thermal behaviour can be improved if high thermal conductive materials (e.g. insulated metal substrate, potting material) are employed combined with proper layout of power dissipating components. Advanced materials will increase the overall costs but reduced internal temperatures of the components promise longer lifetime of the whole MIC - therefore ecient utilisation of the whole PV system and costs savings as a consequence. Following these criteria, a 1st generation Low Prole MIC was designed [6]. A comparison of power densities of MICs in Fig. 6 shows that this MIC prototype has twice the power density and almost half of the height of commercially available MIC. This Low Prole MIC can be mounted easily to the frame of the solar module and due to the reduced weight, the mechanical connection is more reliable and cheaper. Design and packaging steps for the 1st generation Low Prole MIC will be explained in the following subsections. 3.1. Structure and mode of operation of the MIC The block diagram in Fig. 7 illustrates the structure and components of the MIC. The topology used in the

Fig. 7. Schematic of the Module Integrated Converter.

M. Meinhardt et al. / Microelectronics Reliability 39 (1999) 14611472

1465

Fig. 8. Side-view of the 1st generation prototype of a Module Integrated Converter for PV application (height 1.5 cm).

design of the MIC has very good eciency in partial load range. It has been selected out of 36 available topologies, which are discussed in [7]. The solar module on the left generates a maximum power of 110 W at a voltage between 26 V and 37 V. The DC-lter capacitor (1) is used to lter the 100 Hz pulsation of the power typical for single-phase applications. The resonant converter (2) consists of a single halfbridge, resonant circuit and transformer. In order to increase the eciency and decrease the required size of the converter the resonant inductance and capacitance are integrated in the transformer and the halfbridge, respectively. The switching frequency of converter Mosfets is up to 500 kHz. The controlled rectier (3) is realised as diode halfbridges which are alternatively switched on and o by Mosfets switching with 50 Hz. The AC lter (4) ensures compatibility with the EN standards. The control unit (5) includes the fast control and trigger equipment for the resonant converter, the control for the rectier and the supervisory control, which is responsible for maximum-power-point tracking and the unity power factor operation of the MIC. Also some safety issues like the monitoring of the

MIC's internal temperature are carried out by the control. 3.2. Internal packaging concept of the MIC The physical realisation of the electrical circuit and the arrangement of the main components in the 1st generation Low Prole prototype of a MIC are shown in Figs. 810. The side view in Fig. 8 shows that circuitry of the MIC is spread over two circuit boards. The lower board (shown in Fig. 9) is the Insulated Metal Substrate (IMS) with a single conduction layer, containing all power dissipating components such as Mosfets and diodes as well as the magnetic components. The upper substrate is realised as a four layer Printed Circuit Board made of FR4. The FR4 board (shown in Fig. 10) contains the windings of the transformer and the inductors of the AC line lter as well as the control and measurement circuitry. Throughhole components such as low prole DC capacitors are also connected to the FR4 board. This ``Double Decker'' design (Fig. 10) leads to a very good thermal behaviour of the MIC due to the excellent thermal

Fig. 9. Top-view of IMS baseplate of the 1st generation prototype of the MIC (upper FR4-PCB removed).

Fig. 10. Top view of the 1st generation prototype of the MIC (width 13 cm, length 9.5 cm).

1466

M. Meinhardt et al. / Microelectronics Reliability 39 (1999) 14611472

Fig. 11. Exterior packaging concept for the MIC.

conductivity of IMS, combined with the opportunity to build the control circuit on both sides of a four layer PCB and to integrate the windings of the planar magnetic devices into this PCB. Two rows of pins are used for the electrical and mechanical connection of the upper FR4 board at a clearance of 7.5 mm above the lower IMS board. This height is determined by the size of power components and control unit between the IMS and FR4 boards. The minimum height of the MIC depends on the height of planar magnetic components (basically the magnetic cores) and the height of the low prole DC capacitors. The 1st generation Low Prole prototype of a MIC is 1.5 cm in height. This was achieved with the electronic components recently available on the market. 3.3. Exterior packaging Fig. 11 shows the schematic of the exterior packaging concept of the MIC. The assembled MIC is covered with a plastic lid and encapsulated with thermally conductive potting material. Packaging for outdoor application is often rated with IP65, which describes an enclosure totally protected against dust and protected from low-pressure water jets from all directions. This IP rating [8] applies to all exterior components potting lid, cables, grommets, connectors etc. that provide protection from external environment (Fig. 12). The adhesive material will provide good adhesion at the joint between the solder mask of the IMS substrate and the potting lid. This will ensure protection against moisture ingress through the joint into the interior of the MIC.

The use of encapsulation has a number of advantages. It provides global protection from the surrounding environment: chemical (moisture etc.), thermal (temperature extremes) and mechanical (dust etc.) as well as being used to improve thermal performance of the MIC. Potting material (encapsulant) should be exible to put minimum thermo-mechanical stress on the internal electronic components. It also has to ensure uniform heat distribution in the bulk due to power losses in the components i.e. eliminate hot spots. More detailed investigation of the potting materials and their impact on the reliability of the MIC is presented in [9]. As a result, silicone rubbers and exible epoxies with medium thermal conductivity are recommended. Further information can only be obtained

Fig. 12. Exterior components (IP 65) of the MIC.

M. Meinhardt et al. / Microelectronics Reliability 39 (1999) 14611472

1467

Fig. 13. MIC mounted on the frame of solar module.

by applying thermo-mechanical modelling and adequate simulation. To ensure better thermal behaviour, the MIC should be mounted with at least one side of the IMS to the aluminium frame of the solar module. The nal height of the enclosed converter does not exceed 1.7 cm, which is sucient for mounting purpose (see Fig. 13).

by convection and radiation is taken into account by assigning a constant heat ux value through the surface of the MIC. As thermal behaviour of the MIC is mainly conductive, the temperature inside the MIC has nearly linear dependency on the frame temperature. That means the frame temperature is only an oset value for the component's temperature. 4.1. Thermal evaluation of a solar module For the detailed simulation of the internal temperature of the MIC, a simplied thermal model of the solar module can be used. These investigations showed that the heat ux running from a MIC mounted to the frame of the solar module has a negligible eect on the frame's temperature. Therefore the frame of the solar module on which the MIC is mounted can be modelled by a heat sink with constant temperature. The temperature of the frame of the solar module depends on the method of mounting the solar module (e.g. thermal resistance of solar module stand or mounting structure). Since the eciency of solar modules depends very much on the temperature of the solar cells, measurements presented in the literature always refer to the solar cells' temperature measured at the back of the solar module. Fig. 14 shows temperature frequency

4. Thermal simulation of the MIC The lifetime of electronic components is mainly determined by their temperature. Thermal simulations have been carried out to determine the inuence of dierent design possibilities according to their impact on the resulting temperature of these crucial components. The results of thermal simulation combined with the failure rate calculation provide us with the information about the lifetime (and reliability) of the MIC. The design option, which showed the lowest failure rate, is called ``optimised design'' in the following. Thermal analysis of solar module and MIC (modelled as a box) showed that the heat dissipated by convection and radiation is about 10% of the overall heat. Therefore all simulations described in the following are based on conductive heat transfer. The heat dissipation

Fig. 14. Temperature frequency distribution of solar cells measured on the back of a solar module during one year of operation.

1468

M. Meinhardt et al. / Microelectronics Reliability 39 (1999) 14611472

Fig. 15. Cross-sections of the capacitor model.

distribution for solar cells during one year of operation. This frequency distribution is based on measurements taken in Hanover and Munich published in [10] and [11]. Measurements of the frame temperature are not available. However, the frame temperature will always be lower than the solar cells' temperature. Therefore a maximum solar module frame temperature of 758C is taken as a worst case estimation based on measurements presented in [12]. 4.2. Thermal model of the MIC The MIC is simulated with the thermal simulation program Flotherm by Flomerics. The model is based on the physical structure of the MIC described in section 3.1. The only components modelled on the IMS layer are those which produce signicant heat as well as the constructive components which direct the heat ux (e.g. the pins). The drivers and measurement circuitry as well as the terminals were not included. 4.2.1. DC-lter capacitors The most important part, the capacitors, are modelled as shown in Fig. 15. Since the program Flotherm is not capable of modelling circular components, the capacitors are cuboid. Two resistive and one very conductive layer simulate that the capacitors have a direct connection to the IMS only along one line. Modelling the inside of the capacitors is based on a cross-section of a real one. The aluminium case is covered with a plastic layer except one end. The surface of this square model is about 30% larger than the real capacitor. However, this has little impact due to the very low thermal conductivity of the plastic layer. The terminals of the capacitors have been neglected. 4.2.2. Lower IMS layer, Mosfets and pins The IMS is modelled as an aluminium block with a thin dielectric layer on top of it. The Mosfets have the same overall size and, more importantly, the same thermal resistance between the junction and the case as

in the data sheets. They produce the losses in the circuitry. The rows of pins are modelled as walls with an equivalent thermal conductivity. 4.2.3. Transformer The transformer consists of an EE-core, primary and secondary windings. The core is modelled with ve cuboid blocks. Each block produces the same heat per volume. The windings produce the heat equal to their copper losses. 4.2.4. Upper FR4 board, control unit and inductors The FR4 Printed Circuit Board is thermally anisotropic and has an in-plane conductivity, which depends on the number of layers and thickness of the copper tracks. This conductivity can be up to twenty times as high as the nominal conductivity. Therefore it is modelled as a cuboid block with the high in-plane conductivity and a resistive layer in the middle to reduce the normal conductivity. The cores of the inductors on the top of the FR4 are modelled in the same way as the transformer. The inductor windings produce an amount of heat equal to the copper losses. The control circuitry is a thin cuboid block, which produces the losses equally distributed over the area used by the control components. 4.3. Internal temperature of the ``optimised design'' MIC The results of the 3D thermal simulation of an ``optimised packaging design'' for the MIC are presented in Fig. 16. It shows the hot spot temperature distribution of the components crucial to reliability issue and control circuitry (including temperature sense). The MIC is assumed to be operating in rated mode i.e. temperature of the frame of solar module 758C, DC-power 110 W, MIC eciency 87%. The hot spot temperatures of the components are determined using 2D plots of temperature distribution (Fig. 17). It

M. Meinhardt et al. / Microelectronics Reliability 39 (1999) 14611472

1469

Fig. 16. Temperature distribution at the surface of the MIC.

can be seen from the diagram in Fig. 18 that all reliability crucial components especially the electrolytic DC capacitors have a very small temperature rise (less than 11 K). For operation at rated power the temperature rise of the transformer is 7.5 K. The hottest spot of the entire MIC is the core of the AC inductor lters with a temperature rise of 12 K. The ``optimised design'' employs a potting material with a thermal conductivity of 0.51 W/(m K) to achieve a good heat transfer between FR4 and IMS which leads to a low temperature of the control circuitry on the FR4. Encapsulation materials with a good moisture resistance usually have a lower thermal conductivity (down to 0.03 W/(m K)). For that case where the nal design of the MIC demands a moisture resistant encapsulant, the results in Fig. 19 show that the impact on the crucial components is marginal if the

material with thermal conductivity of 0.5 W/(m K) is used. The mounting position of the converter on the solar module has also a large impact on the internal temperature of the MIC. Fig. 20 shows the optimum position for mounting the MIC on the frame of the solar module and two alternative possibilities (``Variant Capacitor'' and ``Variant Mosfet''). If the MIC is mounted to the frame in the ``Variant Capacitor'' position the DC-capacitors are very close to the frame, which operates as a heat sink. But the results of the thermal simulation illustrated in Fig. 19 show that for mounting position ``Variant Capacitor'' the DC-capacitors are hotter compared to the optimum position. The reason for that is the heat ux running from the Mosfets towards the frame of the solar module. Simulation results in Fig. 19 show a decreased temperature of the rectier Mosfets, if the MIC is mounted to the frame in the ``Variant Mosfet'' position, where the rectier Mosfets are very close to the

Fig. 17. Temperature distribution inside the MIC at the junction temperature of the Mosfets.

Fig. 18. Hot spot temperature of the components (TFrame=758C) and temperature dierence to solar module frame for operation at the rated power and half power.

1470

M. Meinhardt et al. / Microelectronics Reliability 39 (1999) 14611472

Fig. 19. Impact of dierent design features on the component temperature in relation to the ``optimised design''.

frame. On the other hand the temperature of the DC capacitors is increased due to the larger thermal resistance. The optimum position of the MIC on the frame is such that the Mosfets and the DC capacitors are thermally in parallel. In this case capacitors are very close to the frame and the heat ux from the transformer, inductors and Mosfets does not ow through the capacitors. The row of pins connecting FR4 and IMS boards also acts as a shield for heat as it is located in between DC capacitors and power dissipating components. Apart from these design possibilities, it is also possible to increase the lifetime of a MIC by reducing the converted power at very high ambient temperatures. This leads to lower losses, reduced temperature rise (see Fig. 18) and thus to a higher lifetime of the MIC. For this purpose the internal temperature of the MIC should be monitored.

5. ``Optimised design'' vs commercial To assess the improvements achieved by the ``optimised design'' of the MIC, a commercially available converter has been simulated as well. The model used is shown in Fig. 21. The main dierences between both designs are the increased height of the commercially available MIC (3 cm instead of 1.5 cm) and the application of a FR4 substrate (instead of IMS) for attaching the heat dissipating components (Mosfets, transformer). For the comparison of the failure rates of dierent designs the same converter topology was considered (see Fig. 7 in subsection 3.1). Parameters such as encapsulation material or orientation are the same as in the ``optimised design''. The heat transfer through the commercial converter box is very good due to the high thermal conductivity of the aluminium case. The problem occurs in the heat transfer from the Mosfets through the FR4 to the box because of the very poor thermal path between FR4 and box resulting from the low thermal conductivity of FR4. According to Fig. 19, the components of the commercially available MIC show a temperature increase of 1020 K in

Fig. 20. Dierent frame positions for MIC mounting.

Fig. 21. MIC design based on a simplied structure of a commercially available MIC (with FR4 base plate).

M. Meinhardt et al. / Microelectronics Reliability 39 (1999) 14611472

1471

Fig. 22. Impact of dierent design features on the failure rate of the MIC (failure rate of MIC with ``optimised design'': 35.7 failures/106 h operation time).

relation to the ``optimised design''. This corresponds to the temperature values that can be found in [13]. The main motivation for the detailed thermal simulation was to nd a MIC design which is optimised for improved reliability. The failure rate has been calculated based on the procedure described in Military Handbook 217E [2]. The results are rated to the failure rate of the ``optimised design''. According to Fig. 22, the MIC with ``optimised design'' shows the minimum failure rate with the lifetime approx. 70% longer than for the design similar to a commercial product. And even higher costs for low prole electronic components and IMS substrate enable certain savings because of the longer lifetime of the Low Prole MIC. To get more detailed information on the costs and manufacturability further analysis is needed possibly by co-operation with MIC manufacturers.

The design of a MIC was optimised according to reliability issues. This was achieved by a detailed 3D thermal simulation of the MIC and the solar module. The lifetime of the MIC with ``optimised design'' is increased by approx. 70% to compare with commercial product. Future investigations will compare the simulation results with measurements of the MIC's internal temperature. Further steps will be taken to miniaturise the magnetic components. Noise and parasitics will be considered due to the height adjustment. The dierent design possibilities presented in the paper will be compared according to cost and manufacturability issues.

Acknowledgements This work was done in the project called Highly Integrated Converters for Advanced AC-Photovoltaics (HICAAP) which was funded by the European Commision in the framework of the JOULE III Programme (contract number JOR3-CT97-0148). The authors would like to thank all members of the HICAAP working group at the following institutions very much for their contribution to this paper: Soltec, Kapeldreef 60, 3001 Leuven, Belgium; IMEC, Kapeldreef 75, 3001 Leuven, Belgium; ECN, Westerduinweg 3, 1755 ZG Petten, The Netherlands; ISET, Koenigstor 59, 34199 Kassel, Germany. Special thanks to all members of the PEI Technologies Group

6. Conclusions and further work The presented Low Prole design of a Module Integrated Converter (MIC) for Photovoltaic applications has double the power density compared to commercially available MIC. Encapsulation materials have positive impact on the thermal behaviour of the MIC as well as the whole concept of the exterior packaging. The internal temperature of aluminium electrolytic capacitors is the main factor, which inuences their lifetime and the lifetime of the whole MIC.

1472

M. Meinhardt et al. / Microelectronics Reliability 39 (1999) 14611472 parison of dierent topologies according to costs, volume and eciency (in German). In: Proceedings of the 13th Symposium on Photovoltaische Solarenergie, Staelstein, Germany, March, 1998. p. 38690. IEC publication 529 on IP degrees. Leonavicius V, et al. Impact of potting on the reliability of power converters for outdoor applications. In: Proceedings of the IMAPS Europe 99, Harrogate, UK, June 79, 1999. Decker B, Jahn U. Energieertrag von netzgekoppelten PV-Anlagen: Ertragsprognosen und Betriebsergebnisse aus dem 1000-Daecher PV-Programm. In: Proceedings of the 8th Nationales Symposium Photovoltaische Solarenergie, Staelstein, Maerz 1993, 1993. p. 95104. James P. Weather Observations at Munich-Schwabing. Germany, http://www.meteo.physik.uni-muenchen.de/theorie/paul/munich.html, Meteorologisches Institut, Universita t Mu nchen, Theresienstr. 37, 80333 Mu nchen, Germany. Knaupp W. Operational behaviour of roof installed photovoltaic modules. In: Proceedings of the 25th PVSC, Washington, DC, May 1317, 1996. Oldenkamp H, et al. Reliability and accelerated life tests of the AC module mounted OKE4 inverter. In: Proceedings of the 25th PVSC, Washington DC, May 1317, 1996.

at the National Microelectronics Research Centre in Cork, Ireland. References

[1] Development of PV Power Supplies using modular systems design. Report Institute for solar Energieversorgungstechnik e.V., http://www.iset.uni-kassel.de/, Kassel, Germany, 1995. [2] Department of Defence, United States of America Military Handbook 217 E, Reliability Prediction of Electronic Equipment, October 27, 1986. [3] Meinhardt M et al. Reliability of Module Integrated Converters for Photovoltaic Applications. In: PCIM98, Nuremberg, Germany, May 2628, 1998. [4] Evox Rifa, Electrolytic Capacitors, 19971998, Technical Information, Sweden, 1997. [5] Barker Mircrofarads Ic Aluminium Capacitors, Electrolytic Type, Product catalog, USA, 1994. [6] Meinhardt M, et al. Power electronics packaging of ``low prole'' module integrated converters for photovoltaic applications with improved reliability. In: Proceedings of the IMAPS 98, 31st Symposium on Microelectronics, San Diego, November 14, 1998. [7] Myzik J, et al. Module integrated converter a com[8] [9]

[10]

[11]

[12]

[13]